OLIF V2

1
OLIF V2

• Gr. ThurmairApril 2000

2
OLIF Overview

• Rationale
• Principles
• Entries
• Descriptions
• Header
• Examples
• Status

3
The Exchange Problem

• the same text is processed with different tools
• how do these tools communicate?

4
The Resource Problem
Translation Project

• how often will the same term be stored?
• who pays for redundant maintenance?

5
The Migration Problem
use of translation tool 1

• do you have to re-build your language
  resources?
• who pays for rebuilding lexicons and memories?

6
Problems in Exchange

• Different purposes of exchange
• data import / export
• data validation
• Different content of exchange
• terminological data, different for each system
• lexicographical data, different for each system
• Different structures of exchange files
• trend towards markup structures (SGML/XML)

7
OLIF Principles

• Keep it simple!
• flat feature value structures
• standard software environment
• Keep it pragmatic!
• worry only about whats there
• bottom-up compare what systems have

8
OLIF V2 formalism

• Define a representation formalism
• XML
• general format processing tools available
• but overhead in markups gt not suitable for
  very large files
• Coding standard UTF-8
• File structure
• header globals, defaults, definitions
• body sequence of entries

9
OLIF principles

• Entries are concept-based
• i.e. we describe word senses different
  readings -gt different entries
• Entries have (monolingual) descriptions
• both for MT and terminology
• Entries have links to other entries
• inner-language crossreference links
• intra-language transfers (multilingual,
  directed)

10
Entry Structure

• central information
• definition features
• administrative features
• monolingual / linguistic feature set
• terminological feature set
• transfer features
• cross-references / links

11
Definition of the entry

• Obligatory features
• language
• (only language or also locale?)
• canonical form
• do we need guidelines (multiwords)
• part of speech
• only open word classes N, V, A, Adv, Prep
• domain information (semantics)
• gt a common top level classification?
• reading no.

12
Scope of the descriptions

• minimal linguistic descriptions
• features which everybody has / needs
• minimal terminology descriptions
• feature set of e.g. Interval
• minimal transfer descriptions
• equitype, tests, transfers
• minimal thesaurus / ontology relations
• ISO standards
• additional fields for personal use

13
Linguistic Descriptions

• Morphological Features
• entry type
• abbreviation, single word, compound, multiword)
• inflection class
• enumeration of inflection patterns, per category
• gender
• (special) number
• singulare / plurale tantum
• degree / comparative

14
Linguistic descriptions

• Syntactic Features
• Syntactic Type
• (subcategorisation of part-of-speech)
• Syntactic Frame
• Argument structures (DObj, PObj-for, ...)
• (Transitivity)
• intransitive, transitive

15
Linguistic Descriptions

• Semantic Features
• Semantic type
• for subclasses only?
• who has / needs it?

16
Terminological Descriptions

• Minimum needed to validate an entry(Interval)
• Definition
• Context
• Scope
• Comment / Note
• (validation status)
• (a three-level hierarchy)

17
Transfer Descriptions

• Equivalence type
• full - partial (subset / superset) - none
• for reversible entries
• Tests and Actions
• (to be worked out)
• Comment
• (Definition of the target link)

18
Cross-Reference Descriptions

• Linktype
• thesaurus relations
• broader / narrower / synonym / related
• additional customisable relations
• abbreviation_for, forbidden, outdated
• Definition of the target link

19
Administrative Information

• Source of Entry
• string
• Author
• creation author
• last modofication
• Date
• creation date
• last modification date

20
Header Information

• Definition of Encoding
• given in the XML statement (UTF-8)
• Definition of features / values used
• Definition of default values

21
Example (1)
ltENTRYgt ltMONOgt
ltLGgt de lt/LGgt
ltCANgt Brot
lt/CANgt ltCATgt noun lt/CATgt
ltSAgt gv lt/SAgt ....
lt/MONOgt lt/ENTRYgt
22
Example 2a
ltEntrygtltMONOgt.... ltCANgt offshore account
lt/CANgt ltCATgt noun lt/CATgt ltSAgt
Money-Laundering lt/SAgt ...lt/MONOgtltXFRgt ltCANgt
compte en banque à létranger lt/CANgt ltLGgt
fr lt/LGgt ltCATgt noun lt/CATgt ltSAgt
Money-Laundering lt/SAgt lt/XFRgtltXFRgt ltCANgt
Auslandskonto lt/CANgt ltLGgt de lt/LGgt
... lt/Entrygt
23
Example (2b)
ltEntrygtltMONOgt ltCANgt compte en banque à
létranger lt/CANgt ltCATgtnoun lt/CATgt
ltSAgtMoney-Laundering lt/SAgt ...lt/MONOgtltXFRgt
ltCANgtoffshore account lt/CANgt ltLGgten lt/LGgt
ltCATgtnoun ltSAgtMoney-Laundering lt/SAgt
lt/XFRgtltXFRgtltCANgtAuslandskonto lt/CANgt ltLGgtde
lt/LGgt ... lt/Entrygt
24
OLIF Status

• Implementation of a central DB
• Implementation of OLIF parser generatorBUT
• different flavours of OLIF
• dependent on projects (OTELO - Aventinus)
• different formalisms (own, SGML, XML)

25
Status

• Verification of OLIF
• MT lexicons (Logos, T1)
• Term Bases (SAPterm, DanTerm)
• Converter prototypes
• T1 lt-gt OLIF, Logos lt-gt OLIF
• Term Lookup systems
• based on OLIF-type database
• Comparisons with other formats